Energy management and control system for intelligent building

Abstract

The application discloses a kind of wisdom building energy management and control systems, belong to automatic control technical field.The system is in view of the problem that traditional building energy system is in personnel transition period of departure continues high energy consumption operation, by deploying sensor network in specific room and collecting real-time personnel distribution state data;Data processing unit calculates real-time personnel density value, moving average of density value in continuous period and its dispersion accordingly;When detecting that average value of dispersion in predetermined window is lower than first threshold value and density value continues to drop, it is determined that the room will enter low personnel density state;Energy control unit generates first control instruction accordingly, in the transition period after determination to personnel density reaches second threshold value, according to real-time drop rate, the refrigeration or heating temperature set value of climate regulation system in this room is linearly adjusted in stages, when reaching second threshold value, second control instruction is generated to close climate regulation system.The system is used to optimize the energy consumption of intermittent use space between buildings.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology. More specifically, this invention relates to an intelligent building energy management system. Background Technology

[0002] Building energy systems exhibit significant energy waste in intermittently used areas such as conference rooms and lecture halls. These spaces have distinctly time-sensitive population gatherings, with meetings or events typically ending followed by a gradual departure of personnel. Under traditional control strategies, systems often only shut down equipment after detecting complete vacancy, resulting in the climate control system continuing to operate near full capacity during the transition period when population density gradually decreases. This delayed response leads to substantial wasted cooling or heating energy, particularly in scenarios with extended periods of occupancy.

[0003] Existing solutions attempt to optimize control timing through people counting technology, but they still face many challenges in practical applications. Conventional infrared sensing or single-camera solutions struggle to effectively distinguish between brief movements of people and their actual departure, especially when people are in a relatively static posture (such as sitting at an office), where recognition accuracy drops significantly. While carpet-type pressure sensors can detect standing positions, they have a high false positive rate due to pressure from objects. These limitations prevent the system from accurately capturing the initial trends in people density changes, hindering the early intervention of control strategies.

[0004] Furthermore, if the temperature setting is adjusted before the room is completely empty, traditional systems typically use a single-rate, step-like adjustment. This abrupt temperature change can easily cause thermal discomfort for those staying in the room, especially in large spaces where differences in thermal perception between different areas are more pronounced. Although some systems have introduced zone temperature monitoring, their response speed to local hot or cold spots is insufficient, with a delay of tens of seconds between detection and compensation, failing to promptly eliminate discomfort for occupants.

[0005] The underlying reasons for the above problems are: the randomness and complexity of personnel flow patterns, and accurately identifying the starting point of the departure trend requires solving the problem of separating instantaneous fluctuations from the true trend; at the same time, implementing gradual energy-saving adjustments while ensuring the comfort of remaining personnel requires overcoming the technical obstacles of multi-variable coupled control. These factors together lead to the common situation where climate control systems for intermittently used spaces in buildings struggle to balance energy efficiency optimization and comfort assurance during transition periods. Summary of the Invention

[0006] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0007] To achieve these objectives and other advantages according to the present invention, a smart building energy management system is provided, comprising:

[0008] Sensor networks are deployed in specific rooms to collect real-time data on the distribution of people.

[0009] The data processing unit, which is connected to the sensor network, includes a real-time analysis module. The real-time analysis module calculates the current personnel density value of a specific room, the moving average value of the personnel density value within multiple consecutive sampling periods, and its dispersion based on real-time personnel distribution data. When the arithmetic mean of the dispersion within a predetermined continuous time window is detected to be lower than a first threshold and shows a continuous downward trend, it is determined that the specific room is about to enter a low personnel density state.

[0010] An energy control unit is connected to a data processing unit and energy-consuming devices in the room. When it receives a judgment signal from the data processing unit indicating that a specific room is about to enter a low-personnel-density state, it generates a first control command. When the room's personnel density reaches a second threshold, it generates a second control command. The energy-consuming devices include a climate control system.

[0011] The first control command is used to pre-adjust the climate control system of the specific room during the transition period from when the specific room is determined to be about to enter a low-personnel-density state until the personnel density of the specific room reaches a second threshold.

[0012] The second control command is used to turn off the climate control system of the specific room;

[0013] The pre-adjustment includes: linearly increasing the cooling temperature setpoint of the climate control system or linearly decreasing the heating temperature setpoint of the climate control system in stages according to the real-time rate of decrease of the current occupancy density in the room; the rate of linear increase or decrease is in the range of 0.1℃ to 0.5℃ per minute.

[0014] Preferably, the sensor network includes a wide-angle camera deployed at a predetermined height above the ceiling in a specific room and a carpet-like pressure sensor array deployed at a predetermined position on the floor of the specific room. The wide-angle camera has a field of view of not less than 170 degrees and acquires image data including the outline of a person's head at a first sampling frequency. The carpet-like pressure sensor array consists of multiple pressure sensing areas arranged in a grid, with each grid cell having a side length ranging from 15cm to 30cm. The carpet-like pressure sensor array acquires foot pressure distribution data at a second sampling frequency.

[0015] The data processing unit also includes a coordinate calibration module and a state fusion module;

[0016] The coordinate calibration module, based on a predefined three-dimensional spatial coordinate mapping relationship of the room, transforms the head contour coordinates identified in the image data and the foot pressure center coordinates identified in the foot pressure distribution data into two-dimensional plane coordinates in the same room coordinate system, and pairs and associates the two-dimensional coordinates of the head contour and the two-dimensional coordinates of the foot pressure center at the same moment.

[0017] The state fusion module calculates the dwell time of people in each pressure-sensing grid cell based on the successfully paired coordinate data. When the dwell time in a single pressure-sensing grid cell continuously exceeds a preset static determination threshold, it determines that there are stationary people in that grid cell. It aggregates all grid cells that have been determined to have stationary people to generate a two-dimensional location distribution map of people in the specific room. The personnel distribution status data includes the two-dimensional location distribution map of people and the real-time occupancy count of the pressure-sensing grid cells.

[0018] Preferably, the method for the real-time analysis module to calculate the personnel density value in the room at the current moment, calculate the moving average of the personnel density value over a predetermined number of consecutive time sampling periods, and calculate the dispersion of the real-time personnel density value of each sampling point relative to the moving average over the predetermined number of consecutive time sampling periods includes:

[0019] The real-time personnel density value at the current moment is calculated based on the ratio of the total number of pressure sensing grid cells occupied in the current sampling period to the total number of pressure sensing grid cells in the specific room.

[0020] A weighted moving average is calculated for the real-time personnel density value sequence of N consecutive sampling periods, where N is an integer greater than or equal to 10 and less than or equal to 30. In the weighted moving average calculation, the real-time personnel density values ​​corresponding to the most recent m sampling periods are given higher weight coefficients, where m is an integer greater than or equal to 3 and less than or equal to 8, and the weight coefficients corresponding to the sampling periods closer to the current time are larger.

[0021] Calculate the variance between the real-time population density value of each sampling point and its corresponding weighted moving average value within the N consecutive sampling periods, and use the variance as a quantitative indicator of dispersion;

[0022] Preferably, the specific process by which the real-time analysis module determines that the specific room is about to enter a low-personnel-density state includes:

[0023] Linear regression analysis was performed on the real-time personnel density value sequence of the N consecutive sampling periods to obtain the slope value of personnel density changing with time;

[0024] When preset conditions are met simultaneously, it is determined that the specific room is about to enter a low-personnel-density state. The preset conditions include: a) the variance is lower than a preset variance threshold for K consecutive sampling periods, where K is an integer greater than or equal to 3 and less than or equal to 10; b) the slope value is negative; c) the absolute value of the slope value is greater than a preset slope threshold for K consecutive sampling periods; wherein, the variance threshold ranges from 0.5 to 1.5, and the slope threshold ranges from a decrease of 0.05 to 0.2 density units per minute.

[0025] Preferably, the pre-adjustment process of the energy control unit specifically includes:

[0026] The real-time rate of decrease in personnel density, calculated in real time by the data processing unit, is obtained. The real-time rate of decrease in personnel density is the absolute value of the slope value obtained by linear regression analysis.

[0027] The real-time rate of decrease in personnel density is input into a predefined adjustment coefficient calculation function, and a first adjustment coefficient is output. The adjustment coefficient calculation function satisfies the following: when the real-time rate of decrease in personnel density is within the range of a preset lower threshold to an upper threshold, the first adjustment coefficient is directly proportional to the real-time rate of decrease in personnel density, and the proportionality constant ranges from 0.5 to 2.0.

[0028] Based on the first adjustment coefficient α, calculate the target adjustment range ΔT of the climate control system set temperature during the transition period: ΔT = α × ΔT0, where ΔT0 is the basic adjustment range, which ranges from 1.0℃ to 3.0℃;

[0029] The transition period is divided into two consecutive adjustment sub-stages. In the first adjustment sub-stage, the cooling temperature setpoint or heating temperature setpoint of the climate control system is adjusted in the direction of the target adjustment range at a first temperature change rate. In the second adjustment sub-stage, the setpoint is further adjusted in the direction of the target adjustment range at a second temperature change rate.

[0030] Wherein, the first temperature change rate is less than the second temperature change rate; the first temperature change rate ranges from 0.1℃ to 0.3℃ per minute, and the second temperature change rate ranges from 0.2℃ to 0.5℃ per minute.

[0031] Preferably, the sensor network further includes multiple temperature sensors, multiple humidity sensors, and wind speed sensors of several variable air volume terminal devices integrated into a climate control system, all deployed in a specific room. The multiple temperature sensors are used to collect location temperature data, the multiple humidity sensors are used to collect location humidity data, and the wind speed sensors are used to collect terminal wind speed data.

[0032] The data processing unit has a preset sensor-area mapping relationship: the specific room is divided into several areas, each area corresponds to a temperature sensor, a humidity sensor and a variable air volume terminal device. Each area is numbered, and the temperature sensor, humidity sensor, wind speed sensor and variable air volume terminal device in the area are given the same unique identifier. The identifiers of the temperature sensor, humidity sensor, wind speed sensor and variable air volume terminal device in each area are associated with the number of the area.

[0033] After the energy control unit generates the first control command, the data processing unit receives in real time personnel distribution status data, location temperature data, location humidity data, and terminal wind speed data collected from the sensor network. Based on the personnel distribution status data, it identifies the remaining personnel location areas in the current room. According to the preset sensor-area mapping relationship, it matches the corresponding location temperature data, location humidity data, and terminal wind speed data for each remaining personnel location area. Using the PMV model, it calculates the real-time local thermal comfort value of each remaining personnel location area based on the matched location temperature data, location humidity data, and terminal wind speed data. When any real-time local thermal comfort value exceeds the preset comfort threshold range, it sends a compensation trigger signal and the corresponding target variable air volume terminal device identifier to the energy control unit.

[0034] While the energy control unit outputs the first control command for pre-adjustment, if it receives the compensation trigger signal and the target variable air volume terminal device identifier, it generates a dynamic compensation command. The dynamic compensation command directionally controls the target variable air volume terminal device to increase its air volume while maintaining the set temperature adjustment process. The increase in air volume is: Δ air volume = k × |PMV offset|, where k is a proportional coefficient and PMV offset is the difference between the real-time local thermal comfort value and the comfort threshold boundary value.

[0035] Preferably, the response delay of the air volume adjustment does not exceed 15 seconds.

[0036] Preferably, the data processing unit transmits real-time personnel distribution data and environmental parameters to the sensor network via an RS-485 bus or Ethernet communication protocol.

[0037] The beneficial effects of this invention are mainly reflected in three aspects: improved energy efficiency, guaranteed user comfort, and enhanced system intelligence.

[0038] This system enables early intervention for energy-saving operations by accurately identifying the impending low-density state before people have completely left the room. Specifically, the system continuously collects personnel distribution data using a sensor network, and the real-time analysis module of the data processing unit calculates personnel density, moving average, and dispersion. When the system detects that the average dispersion is below a preset threshold and the personnel density shows a continuous downward trend, it determines that the room is about to enter a low-density state. At this point, the system triggers a pre-adjustment mechanism for a transition period, rather than passively waiting for the room to become completely empty. This significantly extends the energy-saving operation window and avoids the energy waste caused by traditional systems continuously running high-intensity cooling or heating during the departure of personnel.

[0039] The phased linear temperature control strategy in the pre-adjustment stage effectively balances energy-saving goals with the comfort requirements of personnel. The system dynamically calculates and sets the temperature target adjustment range based on the real-time rate of decrease in personnel density. This adjustment process is divided into at least two consecutive sub-stages: the first stage uses a slower rate of temperature change for fine-tuning, while the second stage uses a slightly faster rate to complete the remaining adjustment. This gradual adjustment strategy ensures that when people are still in the room, the temperature changes are gentle and imperceptible, avoiding discomfort caused by sudden temperature changes. Simultaneously, the system combines the base adjustment range with an adjustment coefficient dynamically calculated based on the departure speed, ensuring that the final temperature setpoint adjustment is both energy-saving and reasonable.

[0040] The system further introduces a dynamic compensation mechanism for local thermal comfort, significantly improving the user experience in complex environments. During pre-adjustment, by using temperature, humidity, and wind speed sensors deployed in various areas of the room, combined with a real-time updated location map, the system can accurately identify areas where people are still gathered and match their microenvironmental parameters. Using a PMV model, the system calculates the real-time local thermal comfort value for each area. Once it detects that the comfort value of a certain area exceeds a preset threshold, the system can trigger directional airflow compensation in a very short time. This compensation mechanism alleviates local discomfort by rapidly increasing the airflow of the corresponding variable airflow terminal unit. The compensation magnitude is proportional to the PMV offset, ensuring precise intervention. Most importantly, the response delay of the entire compensation process is strictly controlled within 15 seconds, allowing the system to respond and adjust rapidly as soon as the user perceives slight discomfort, effectively maintaining user satisfaction.

[0041] The reliability of the overall system architecture and the accuracy of data processing form the foundation for effective operation. The sensor network employs a complementary design of wide-angle cameras and a carpet-like array of pressure sensors. Through coordinate calibration and state fusion, it accurately generates multi-dimensional state data, including a map of stationary personnel locations and grid occupancy counts, providing a solid foundation for density calculation and trend analysis. In the data processing stage, a weighted moving average algorithm with higher weights for recent data is used to calculate the moving average, combined with variance quantification of dispersion, effectively smoothing instantaneous fluctuations and improving the accuracy of trend determination and anti-interference capability. When determining if a low-density state is about to be entered, a multi-condition joint mechanism is implemented, requiring variance to be continuously below a threshold, slope to be continuously negative, and the rate of decline to be continuously greater than a threshold, significantly reducing the risk of misjudgment. Furthermore, data transmission between the core sensor network and the data processing unit uses RS-485 bus or Ethernet communication protocols, ensuring high reliability and real-time performance of critical data, avoiding potential interference and delay issues in wireless transmission, and providing strong support for stable system operation and precise control.

[0042] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the intelligent building energy management system described in this invention. Detailed Implementation

[0044] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0045] like Figure 1 As shown, the present invention provides a smart building energy management system, characterized in that it includes: a sensor network, a data processing unit, an energy control unit, and energy consumption equipment. The sensor network includes sensors deployed in specific rooms, and the energy consumption equipment includes a climate control system deployed in the specific rooms.

[0046] The sensor network is used to collect real-time data on the distribution of people in a specific room.

[0047] The data processing unit is communicatively connected to the sensor network. The data processing unit includes a real-time analysis module. The real-time analysis module receives real-time personnel distribution data collected by the sensor network from the specific room. Based on the continuous real-time personnel distribution data, it calculates the real-time personnel density value in the room at the current moment, calculates the moving average of the personnel density value over a predetermined number of consecutive time sampling periods, and calculates the dispersion of the real-time personnel density value of each sampling point relative to the moving average over the predetermined number of consecutive time sampling periods. When it is detected that the arithmetic mean of the dispersion over a predetermined consecutive time window is lower than a preset first threshold and shows a continuous downward trend, it is determined that the specific room is about to enter a low personnel density state.

[0048] The energy control unit is communicatively connected to the data processing unit and the climate control system of the specific room. When the energy control unit receives a determination signal from the data processing unit that the specific room is about to enter a low-person density state, it generates a first control command. When the person density of the specific room reaches a preset second threshold, it generates a second control command.

[0049] The first control command is used to pre-adjust the climate control system of the specific room during the transition period from when the specific room is determined to be about to enter a low-personnel density state until the personnel density of the specific room reaches a preset second threshold. The pre-adjustment includes: linearly increasing the cooling temperature setpoint of the climate control system in stages or linearly decreasing the heating temperature setpoint of the climate control system in stages according to the real-time rate of decrease of the current personnel density in the room; the rate of linear increase or linear decrease is in the range of 0.1℃ to 0.5℃ per minute.

[0050] The second control command is used to turn off the climate control system of the specific room.

[0051] Specifically, this smart building energy management system aims to optimize energy use in specific rooms, especially by intelligently controlling the room's climate control system when people are about to leave or when the density decreases significantly. The system's operation begins with a sensor network. These sensors are deployed in the specific rooms that need to be monitored, and their main task is to continuously collect real-time data on the distribution of people in the rooms, understanding changes in the location and number of people.

[0052] The collected personnel distribution data is transmitted to the data processing unit in real time. Within this unit, the core component is the real-time analysis module. This module receives continuous real-time personnel distribution data streams and performs a series of complex calculations and analyses based on this data. First, it calculates the real-time personnel density value in the room at the current moment, which is typically a value reflecting the degree of crowding in the room (e.g., the proportion of occupied area or the number of people per unit area). Next, the module reviews a predetermined period of time (i.e., multiple consecutive time sampling periods) and calculates the moving average of the personnel density value during this period. The moving average helps smooth out short-term random fluctuations and reveals the overall trend of personnel density. Then, the module calculates the dispersion (e.g., variance) of the real-time personnel density value at each sampling point relative to this moving average, reflecting the recent fluctuations in personnel density. A crucial step is that when the real-time analysis module detects that within a predetermined time window, the average dispersion of the personnel density value is below a preset first threshold (indicating smaller fluctuations, a trend towards stability or regularity), and the personnel density value itself shows a continuous downward trend, it makes a prediction: the specific room is about to enter a low personnel density state. This prediction occurs before the population density actually drops to a very low level (e.g., close to zero).

[0053] Once the data processing unit determines that a "low occupancy density state" is imminent, it immediately sends a judgment signal to the energy control unit. The energy control unit, the system's execution center, connects to the data processing unit and the climate control system of that specific room. Upon receiving the judgment signal, the energy control unit generates a first control command. This command pre-adjusts the climate control system during the transition period between when the room is determined to be entering a low occupancy density state and when the occupancy density actually reaches a preset second threshold (e.g., 0 or close to 0). The core strategy of this pre-adjustment is to linearly adjust the climate control system's set temperature in stages based on the real-time rate of decrease in the room's current occupancy density (i.e., the speed at which people leave). Specifically, if the system is in cooling mode, the cooling temperature setpoint is linearly increased in stages; if it is in heating mode, the heating temperature setpoint is linearly decreased in stages. This adjustment is not a one-step process but rather a gradual, phased adjustment. For example, the adjustment might be divided into two phases: the first phase adjusts at a slower rate (e.g., 0.1°C to 0.2°C per minute), and the second phase continues to adjust at a slightly faster rate (e.g., 0.3°C to 0.5°C per minute). The adjustment rate is strictly controlled within the range of 0.1°C to 0.5°C per minute to ensure that the temperature change is not too drastic and affects the comfort of the remaining people. This gradual temperature control is intended to reduce the energy load of the climate control system in advance as people gradually leave (e.g., allowing the room temperature to rise slowly when cooling and to fall slowly when heating), achieving energy savings while avoiding discomfort caused by shutting down the system before the last person leaves.

[0054] When the data processing unit detects that the occupancy density in the room has actually reached a preset second threshold (for example, the room is empty), it notifies the energy control unit. At this point, the energy control unit generates a second control command. This command is very direct: shut down the climate control system for that specific room. This is the final step in energy saving, completely stopping energy consumption in unoccupied rooms.

[0055] During room use (such as before the end of a meeting room's use), sensors continuously monitor the distribution of people. When the system predicts, by analyzing personnel density data and its changing trends, that all or most people are about to leave (but there are still some people), a pre-regulation mechanism is triggered. The climate control system begins to adjust the set temperature in stages and slowly, based on the speed at which people leave (gradually increasing the set temperature during cooling and gradually decreasing it during heating). For example, if people are leaving at a moderate speed, the set temperature might be increased by 0.2°C per minute (cooling mode). During the time it takes for the last one or a few people to leave the room, the system is already gradually saving energy. When the system confirms that the room is empty (density reaches the second threshold), it immediately shuts off the HVAC system. The main technical advantage of this approach is that it significantly advances the start of energy-saving operations, reducing unnecessary cooling / heating intensity during the transition period before people leave, rather than waiting until everyone has left to shut down the system. It smooths out the temperature change process, ensuring the comfort of remaining people (slow temperature changes) while achieving considerable energy savings, making it particularly suitable for rooms with high personnel turnover (such as meeting rooms and lecture halls). Through intelligent prediction and gradual regulation, the peak energy consumption and total energy consumption of buildings have been effectively reduced.

[0056] Furthermore, the sensor network includes a wide-angle camera deployed at a predetermined height above the ceiling in a specific room and a carpet-like pressure sensor array deployed at a predetermined location on the floor of the specific room. The wide-angle camera has a field of view of not less than 170 degrees and acquires image data including the outline of a person's head at a first sampling frequency. The carpet-like pressure sensor array consists of multiple pressure sensing areas arranged in a grid, with each grid cell having a side length ranging from 15cm to 30cm. The carpet-like pressure sensor array acquires foot pressure distribution data at a second sampling frequency.

[0057] The data processing unit also includes a coordinate calibration module and a state fusion module;

[0058] The coordinate calibration module, based on a predefined three-dimensional spatial coordinate mapping relationship of the room, transforms the head contour coordinates identified in the image data and the foot pressure center coordinates identified in the foot pressure distribution data into two-dimensional plane coordinates in the same room coordinate system, and pairs and associates the two-dimensional coordinates of the head contour and the two-dimensional coordinates of the foot pressure center at the same moment.

[0059] The state fusion module calculates the dwell time of people in each pressure-sensing grid cell based on the successfully paired coordinate data. When the dwell time in a single pressure-sensing grid cell continuously exceeds a preset static determination threshold, it determines that there are stationary people in that grid cell. It aggregates all grid cells that have been determined to have stationary people to generate a two-dimensional location distribution map of people in the specific room. The personnel distribution status data includes the two-dimensional location distribution map of people and the real-time occupancy count of the pressure-sensing grid cells.

[0060] Specifically, the sensor network consists of two main types of devices. The first type is a wide-angle camera installed at a predetermined height above the ceiling in a specific room. A key performance indicator for this camera is its field of view coverage of at least 170 degrees, ensuring coverage of most of the floor area and minimizing blind spots. This camera continuously acquires image data, including the outline of people's heads, at a specific initial sampling frequency (e.g., 5-10 frames per second). The second type of device is a carpet-like array of pressure sensors laid at predetermined locations on the floor of the specific room. This array consists of numerous independent pressure-sensing areas arranged in a regular grid. Each grid cell has a side length between 15cm and 30cm, a size range that effectively captures the footprints of a single person standing or sitting, while balancing cost and accuracy. The pressure sensor array acquires foot pressure distribution data on the room floor at a second sampling frequency (e.g., 2-5 times per second), recording the pressure value changes in each grid cell.

[0061] The acquired raw image data and pressure distribution data are transmitted to the data processing unit for processing. The data processing unit contains two key modules: a coordinate calibration module and a state fusion module.

[0062] The primary task of the coordinate calibration module is to unify raw position data from different sensors and located in different spatial coordinate systems into a single two-dimensional plane coordinate system for the room. This module utilizes a predefined three-dimensional spatial coordinate mapping relationship for the room (usually obtained through calibration during installation) to geometrically transform the center coordinates of the person's head outline identified in the wide-angle camera image (usually represented by image pixel coordinates) to two-dimensional plane coordinates on the room floor. Simultaneously, the module also maps the coordinates of the plantar pressure center points identified by the carpet-type pressure sensor array (usually based on grid cell indexing and pressure center calculation) to the same two-dimensional plane coordinate system for the room floor. After completing the coordinate transformation, the module pairs and associates the two-dimensional coordinates of the head outline and the plantar pressure center acquired at the same time. A typical pairing logic is to find the closest spatially located head and foot point pairs within a reasonable human height range; for example, pressure center points within a reasonable distance (e.g., 0.8m to 1.2m) below a head are likely to belong to the same person.

[0063] The state fusion module performs deeper analysis based on the data successfully paired by the coordinate calibration module. This module calculates the cumulative duration of time a person is identified as residing within each pressure-sensing grid cell (i.e., effective pressure is detected and head vertices are successfully paired). A preset static determination threshold is set (e.g., 30 seconds or 60 seconds consecutively). If the dwell time within a grid cell continuously exceeds this threshold, it is determined that a stationary person exists within that grid cell (e.g., someone sitting at an office or standing still for an extended period). The module aggregates information from all grid cells identified as having stationary persons, ultimately generating a two-dimensional personnel location distribution map reflecting the static distribution of people in a specific room. This map clearly indicates which locations in the room have people staying for extended periods. Simultaneously, the state fusion module also counts the total number of effectively occupied pressure-sensing grid cells within the current sampling period (regardless of dwell time). Therefore, the final generated personnel distribution state data contains two core elements: a two-dimensional personnel location distribution map reflecting the static location distribution of people, and a real-time occupancy count of pressure-sensing grid cells reflecting the overall occupancy of the room (i.e., the total number of occupied grid cells).

[0064] During room use, a wide-angle camera continuously captures overhead images, while a carpet-like pressure sensor continuously senses ground pressure. The data processing unit performs coordinate transformations, pairing and association, dwell time calculations, and stationary person identification every second. For example, in a conference room scenario, when someone enters the room, their head outline and foot pressure are captured and paired. When the person is standing or moving, their dwell time in the corresponding grid cell may be short; when the person sits down to begin a meeting, the dwell time in their grid cell continuously increases. Once it exceeds the stationary identification threshold, the grid cell is marked as having a stationary person and displayed on the location distribution map. Simultaneously, the system counts the total number of occupied grid cells in real time to calculate personnel density. The technical advantage of this solution is that it provides high-precision and multi-dimensional information on room personnel status. By combining image and pressure data, and through coordinate calibration and pairing, it significantly improves the accuracy of personnel location positioning and reduces misjudgments by a single sensor (e.g., cameras cannot identify people bending over or obstructing their view, and pressure sensors cannot distinguish between objects pressing down on them and people standing). In particular, by analyzing dwell time to identify stationary individuals, the generated two-dimensional location distribution map more accurately reflects the actual locations where people are staying (rather than instantaneous passing points). This is crucial for subsequent judgments about whether people are leaving or whether a room is about to become vacant, providing a reliable data foundation for the aforementioned accurate prediction of "entering a low-density state." Simultaneously, real-time occupancy counting provides a direct basis for calculating personnel density.

[0065] Furthermore, the real-time analysis module calculates the personnel density value in the room at the current moment, calculates the moving average of the personnel density value over a predetermined number of consecutive time sampling periods, and calculates the dispersion of the real-time personnel density value of each sampling point relative to the moving average over the predetermined number of consecutive time sampling periods, including the following methods:

[0066] The real-time personnel density value at the current moment is calculated based on the ratio of the total number of pressure sensing grid cells occupied in the current sampling period to the total number of pressure sensing grid cells in the specific room.

[0067] A weighted moving average is calculated for the real-time personnel density value sequence of N consecutive sampling periods, where N is an integer greater than or equal to 10 and less than or equal to 30. In the weighted moving average calculation, the real-time personnel density values ​​corresponding to the most recent m sampling periods are given higher weight coefficients, where m is an integer greater than or equal to 3 and less than or equal to 8, and the weight coefficients corresponding to the sampling periods closer to the current time are larger.

[0068] Calculate the variance between the real-time population density value of each sampling point and its corresponding weighted moving average value within the N consecutive sampling periods, and use the variance as a quantitative indicator of dispersion.

[0069] Specifically, the personnel density value is calculated directly based on data collected by the carpet-like pressure sensor array. The real-time analysis module of the data processing unit counts the total number of occupied pressure-sensing grid cells at each sampling period (e.g., every 6 or 10 seconds). This occupied number refers to those grid cells that have detected effective pressure and are determined to be occupied by personnel. Simultaneously, the system knows the total number of pressure-sensing grid cells laid in that specific room. The real-time analysis module calculates the ratio of the current total number of occupied grid cells to the total number of grid cells in the room to obtain the real-time personnel density value (D) at the current sampling time. current For example, if a conference room has 100 grid cells, and 30 grid cells are currently occupied, then the real-time personnel density value D is... current = 30 / 100 = 0.3 (or 30%).

[0070] To analyze trends in population density and smooth short-term fluctuations, the real-time analysis module performs a weighted moving average (WMA) calculation on the real-time population density value sequence over multiple consecutive sampling periods. Here, the number of consecutive periods, N, is a preset integer ranging from 10 to 30 (e.g., N = 20 periods). The key to the moving average calculation is assigning different weights to data from different periods. Specifically, this method assigns higher weight coefficients to the real-time population density values ​​corresponding to the most recent m sampling periods, where m is also a preset integer ranging from 3 to 8 (e.g., m = 5 periods). Furthermore, the weight coefficients are designed according to the principle that the sampling period closer to the current time has a larger weight coefficient. For example, the first most recent period has the highest weight, followed by the second most recent period, and so on, with the m-th most recent period having a higher weight than earlier periods. This weight allocation method makes the weighted moving average (WMA)... N It is more likely to reflect the latest trends in population density, rather than being overly influenced by earlier, potentially outdated data.

[0071] Dispersion is a key indicator for measuring the degree to which real-time population density values ​​fluctuate around their moving average, and variance is used as the quantitative indicator of this dispersion. The specific calculation process is as follows: For the same set of N sampling periods continuously performing WMA calculations, the real-time analysis module calculates the real-time population density value (Dm) for each sampling point (i.e., each period) in turn. t The weighted moving average (WMA) corresponding to that sampling point t The squared deviation between (D) and (i.e., (D) t - WMA t ) 2 Then, the average of these N squared deviations is calculated to obtain the variance (V) for that time period. ar The variance value directly reflects the fluctuation of population density: a high variance value indicates that the population density fluctuates drastically and is unstable; a low variance value indicates that the population density changes relatively smoothly or has a clear trend.

[0072] During system operation, for example, during the end of a 2-3 minute meeting, the pressure sensor array continuously reports grid occupancy status (e.g., 35 grids occupied in the 1st second, 32 grids occupied in the 6th second, 28 grids occupied in the 12th second, etc.). The data processing unit calculates the real-time density (e.g., D) every sampling period (e.g., 6 seconds). t =0.35, 0.32, 0.28...). Simultaneously, it maintains a window containing the most recent 20 (N=20) density values ​​and calculates their weighted moving average (e.g., the 5 most recent values ​​with the highest weights). Then, it calculates the density value (D...) for each density value in the current window. t ) and its corresponding moving average (WMA) tThe variance of the population density trend is used to determine the stability of the population density trend. This method provides an objective basis for judging the stability of the trend. By quantifying dispersion through variance, the system can distinguish whether people are moving randomly and irregularly (leading to high variance) or leaving in an organized and stable manner (leading to low variance). Combined with the judgment of a downward trend, this weighted moving average method, along with variance calculation, significantly improves the accuracy and reliability of the system's prediction of the key turning point of "approaching a low population density state," avoiding false triggers caused by short-term fluctuations or random movements of a few people, and laying a solid decision-making foundation for subsequent energy-saving pre-adjustment. The weighted moving average assigns higher weights to recent data, making trend judgment more sensitive and enabling faster response to changes in population density where people begin to leave.

[0073] Furthermore, the specific process by which the real-time analysis module determines that a particular room is about to enter a low-personnel-density state includes:

[0074] Linear regression analysis was performed on the real-time personnel density value sequence of the N consecutive sampling periods to obtain the slope value of personnel density changing with time;

[0075] When preset conditions are met simultaneously, it is determined that the specific room is about to enter a low-personnel-density state. The preset conditions include: a) the variance is lower than a preset variance threshold for K consecutive sampling periods, where K is an integer greater than or equal to 3 and less than or equal to 10; b) the slope value is negative; c) the absolute value of the slope value is greater than a preset slope threshold for K consecutive sampling periods; wherein, the variance threshold ranges from 0.5 to 1.5, and the slope threshold ranges from a decrease of 0.05 to 0.2 density units per minute.

[0076] Specifically, the real-time analysis module performs linear regression analysis on the real-time population density value sequence over N consecutive sampling periods. This analysis uses mathematical methods to fit an optimal straight line to describe the overall trend of population density changes over time. The analysis result outputs a slope value, which precisely quantifies the average rate and direction of population density change. For example, a negative slope value clearly indicates that population density is decreasing, while the absolute value of the slope directly reflects the speed of the density decrease.

[0077] Meanwhile, the real-time analysis module continuously monitors the calculated variance index. This variance value quantifies the dispersion of the real-time population density value of each sampling point relative to its corresponding weighted moving average over N consecutive sampling periods. Lower variance indicates that population density changes are stable and have little fluctuation; higher variance means that changes are drastic or irregular.

[0078] Determining whether a "low-density state is imminent" requires three preset conditions to be met simultaneously for K consecutive sampling periods. The first condition is that the variance is below a preset variance threshold for K consecutive sampling periods. This variance threshold is set between 0.5 and 1.5, and the specific value can be optimized and adjusted based on room characteristics and historical data. Meeting this condition indicates that the change in population density is stable and regular, rather than random fluctuations. For example, when a meeting ends in an orderly manner and people leave, the density decrease is usually smooth, and the variance will be below the threshold; conversely, if people are just moving around randomly, the variance may be higher. The second condition is that the slope value obtained from the linear regression analysis must be negative. This directly confirms that the population density is in a continuous downward trend, meeting the basic directional requirement of "imminently entering a low-density state." The third condition is that the absolute value of the slope value must be greater than a preset slope threshold for K consecutive sampling periods. The slope threshold is set to decrease by 0.05 to 0.2 density units per minute. This condition ensures that the rate of decrease in population density is sufficiently significant, not a slow or negligible change. If the absolute value of the slope is greater than the threshold, it means that people are leaving the room at a relatively fast, quantifiable rate.

[0079] Only when all three conditions mentioned above—low variance, negative slope, and a sufficiently large rate of descent—are met within K consecutive sampling periods will the real-time analysis module ultimately determine that the specific room is about to enter a low-occupancy-density state. This determination is the key signal that triggers subsequent energy-saving pre-adjustment operations.

[0080] Towards the end of room usage, for example, about 10 minutes before the end of a meeting, people begin to leave. The system calculates the density value, weighted moving average, variance, and regression slope in each sampling period (e.g., every 6 seconds). Assuming K = 5 periods, a variance threshold of 1.0, and a slope threshold of 0.1 density units per minute, the system determines the room is about to be vacant when it detects that the variance is below 1.0 for 5 consecutive periods (indicating an orderly and stable departure process) and the slope is negative for 5 consecutive periods with an absolute value greater than 0.1 (indicating a sufficiently fast departure speed). This multi-condition joint judgment technique significantly improves prediction accuracy and system robustness. It effectively avoids false triggers caused by brief data fluctuations (such as a brief interruption of the downward trend or a sudden change in speed), ensuring that pre-adjustment energy-saving measures are only activated when people are indeed in a stable and rapid departure state. Simultaneously, by requiring the condition to be met for K consecutive periods, the system can filter out interference from single-point anomalous data. This rigorous judgment logic provides a reliable basis for energy-saving pre-adjustment during the transition period, ensuring that energy-saving opportunities are not missed and that environmental controls are not inappropriately adjusted before the activity has truly ended.

[0081] Furthermore, the pre-adjustment process of the energy control unit specifically includes:

[0082] The real-time rate of decrease in personnel density, calculated in real time by the data processing unit, is obtained. The real-time rate of decrease in personnel density is the absolute value of the slope value obtained by linear regression analysis.

[0083] The real-time rate of decrease in personnel density is input into a predefined adjustment coefficient calculation function, and a first adjustment coefficient is output. The adjustment coefficient calculation function satisfies the following: when the real-time rate of decrease in personnel density is within the range of a preset lower threshold to an upper threshold, the first adjustment coefficient is directly proportional to the real-time rate of decrease in personnel density, and the proportionality constant ranges from 0.5 to 2.0.

[0084] Based on the first adjustment coefficient α, calculate the target adjustment range ΔT of the climate control system set temperature during the transition period: ΔT = α × ΔT0, where ΔT0 is the basic adjustment range, which ranges from 1.0℃ to 3.0℃;

[0085] The transition period is divided into two consecutive adjustment sub-stages. In the first adjustment sub-stage, the cooling temperature setpoint or heating temperature setpoint of the climate control system is adjusted in the direction of the target adjustment range at a first temperature change rate. In the second adjustment sub-stage, the setpoint is further adjusted in the direction of the target adjustment range at a second temperature change rate.

[0086] Wherein, the first temperature change rate is less than the second temperature change rate; the first temperature change rate ranges from 0.1℃ to 0.3℃ per minute, and the second temperature change rate ranges from 0.2℃ to 0.5℃ per minute.

[0087] Specifically, the predefined adjustment coefficient calculation function is a piecewise linear function, the input variable of the piecewise linear function is the real-time decrease rate v, the output variable of the piecewise linear function is the first adjustment coefficient α, and the piecewise boundary of the piecewise linear function is dynamically determined based on the ratio of the volume of the specific room to the nominal cooling capacity, specifically defined as:

[0088] 1. When v≤v min When: α = k1⋅v, where k1 is the first proportionality coefficient, with a value ranging from 0.5 to 0.8;

[0089] 2. When v min <v≤v mid Time: α=β⋅(k2⋅v+b), where k2 is the second proportionality coefficient, ranging from 0.8 to 1.2, b is the bias constant, ranging from -0.1 to 0.1, and β is the dynamic weighting factor, calculated based on the difference ΔT between the current outdoor temperature and the target room temperature: β=1+γ⋅|ΔT|, where γ is the thermal inertia coefficient, ranging from 0.01 to 0.03 / ℃;

[0090] 3. When v > v mid Time: α = min(α) max ,k3⋅v), where k3 is the third proportionality coefficient, ranging from 1.2 to 1.5, and α max The upper limit of the adjustment coefficient is set, with a value ranging from 1.8 to 2.2.

[0091] The segment boundary parameters are calculated according to the following formula: v min =η⋅(V room / Q nom ), v mid =v min +δ, where:

[0092] Q nom : Nominal cooling capacity (kW) of the specific room climate control system;

[0093] V room The volume (m³) of the specific room 3 );

[0094] η: Capacity conversion factor, ranging from 0.02 to 0.05 (min·m) 3 ) / kW;

[0095] δ: Rate interval constant, ranging from 0.03 to 0.07 min. -1 .

[0096] After the pre-adjustment begins, the energy control unit first acquires the real-time rate of decrease in occupant density calculated by the data processing unit. This rate value is directly derived from the absolute value of the slope obtained from linear regression analysis. This value quantifies the average rate of change in occupant density within the room under the current downward trend. For example, if the regression slope is a decrease of 0.15 density units per minute, then the real-time rate of decrease is 0.15.

[0097] Subsequently, the energy control unit inputs this real-time descent rate into a predefined adjustment coefficient calculation function. This function is designed according to specific logical rules: when the real-time descent rate falls between a preset lower threshold (e.g., 0.05 units per minute) and an upper threshold (e.g., 0.2 units per minute), the first adjustment coefficient α output by the function is directly proportional to the input real-time descent rate. This means that the faster the person leaves, the larger the calculated adjustment coefficient. The proportionality constant is preset, with a value ranging from 0.5 to 2.0. For example, if the proportionality constant is 1.0 and the real-time descent rate is 0.15, then the calculated α = 1.5.

[0098] Next, the energy control unit calculates the target adjustment range ΔT that the climate control system's set temperature needs to reach during the transition period, based on the obtained first adjustment coefficient α and a preset base adjustment range ΔT0. The calculation follows the formula: ΔT = α × ΔT0. The base adjustment range ΔT0 is also a preset value, ranging from 1.0℃ to 3.0℃. Continuing with the previous example, if ΔT0 is set to 2.0℃ and α = 1.5, then the target adjustment range ΔT = 1.5 × 2.0℃ = 3.0℃. This means that during the transition period before personnel completely leave, the cooling set temperature needs to be ultimately increased by 3.0℃ (or the heating set temperature needs to be ultimately decreased by 3.0℃).

[0099] To achieve the target adjustment range ΔT while ensuring comfortable temperature changes, the energy control unit divides the entire transition period into at least two consecutive adjustment sub-stages, using different adjustment rates. The division rule for these two sub-stages can be set such that the second adjustment sub-stage begins when the final temperature of the first adjustment sub-stage is 3°C to 5°C away from the set temperature. In the first adjustment sub-stage, the system begins adjusting towards the target adjustment range (i.e., increasing the set temperature during cooling and decreasing it during heating) at a slower first temperature change rate. The first temperature change rate is set to a range of 0.1°C to 0.3°C per minute. For example, an increase of 0.2°C per minute (cooling mode) could be selected. After a period of time, the system enters the second adjustment sub-stage, where a faster second temperature change rate continues to adjust the set temperature towards the target adjustment range. The second temperature change rate is set to a range of 0.2°C to 0.5°C per minute and must be greater than the first temperature change rate. For example, in the second sub-stage, the rate might increase to 0.4°C per minute (cooling mode). This phased adjustment strategy ensures a smooth and slow temperature change in the early stages, followed by a slightly faster adjustment in the later stages.

[0100] After the system determines that the room is about to enter a low-density state and triggers pre-adjustment, for example, when a meeting is ending and people are steadily leaving at a rate of 0.15 density units per minute, the system calculates an adjustment coefficient α = 1.5 (assuming a proportionality constant of 1.0) and a target adjustment range ΔT = 3.0℃ (assuming ΔT0 = 2.0℃). The entire transition period may last 10 minutes. In the first 5 minutes (first stage), the HVAC cooling set temperature is slowly increased by 1.0℃ at a rate of 0.2℃ / minute. In the next 5 minutes (second stage), the set temperature is then increased by an accelerated rate of 2.0℃ at a rate of 0.4℃ / minute, ultimately reaching the target increase of 3.0℃. This phased linear adjustment technique optimizes the balance between energy saving and comfort. In the early stages of the transition period, when there are relatively more people in the room, a slow rate of temperature change (first rate) avoids discomfort for the remaining people due to rapid temperature changes. In the later stages of the transition period, as the number of people further decreased, the system adopted a slightly faster rate of temperature change (second rate) to continue energy-saving adjustments. This ensured that when rooms were about to be vacant or completely vacant, the climate control system was already operating at a more energy-efficient setpoint, reducing unnecessary cooling or heating loads in advance and accumulating considerable energy-saving benefits during the transition period. At the same time, the entire process maintained the smoothness and predictability of temperature changes.

[0101] Furthermore, the sensor network also includes multiple temperature sensors, multiple humidity sensors, and wind speed sensors of several variable air volume terminal devices integrated into the climate control system, all deployed in a specific room. The multiple temperature sensors are used to collect location temperature data, the multiple humidity sensors are used to collect location humidity data, and the wind speed sensors are used to collect terminal wind speed data.

[0102] The data processing unit has a preset sensor-area mapping relationship: the specific room is divided into several areas, each area corresponds to a temperature sensor, a humidity sensor and a variable air volume terminal device. Each area is numbered, and the temperature sensor, humidity sensor, wind speed sensor and variable air volume terminal device in the area are given the same unique identifier. The identifiers of the temperature sensor, humidity sensor, wind speed sensor and variable air volume terminal device in each area are associated with the number of the area.

[0103] After the energy control unit generates the first control command, the data processing unit receives in real time personnel distribution status data, location temperature data, location humidity data, and terminal wind speed data collected from the sensor network. Based on the personnel distribution status data, it identifies the remaining personnel location areas in the current room. According to the preset sensor-area mapping relationship, it matches the corresponding location temperature data, location humidity data, and terminal wind speed data for each remaining personnel location area. Using the PMV model, it calculates the real-time local thermal comfort value of each remaining personnel location area based on the matched location temperature data, location humidity data, and terminal wind speed data. When any real-time local thermal comfort value exceeds the preset comfort threshold range, it sends a compensation trigger signal and the corresponding target variable air volume terminal device identifier to the energy control unit.

[0104] While the energy control unit outputs the first control command for pre-adjustment, if it receives the compensation trigger signal and the target variable air volume terminal device identifier, it generates a dynamic compensation command. The dynamic compensation command directionally controls the target variable air volume terminal device to increase its air volume while maintaining the set temperature adjustment process. The increase in air volume is: Δ air volume = k × |PMV offset|, where k is a proportional coefficient and PMV offset is the difference between the real-time local thermal comfort value and the comfort threshold boundary value.

[0105] Specifically, the sensor network adds three types of sensors to the existing wide-angle cameras and carpet-like pressure sensor array: multiple temperature sensors, multiple humidity sensors, and an air velocity sensor integrated into the variable air volume (VAV) terminal unit of the climate control system. These sensors are strategically deployed in predetermined locations within specific rooms. Multiple temperature sensors collect real-time location temperature data for different areas of the room; multiple humidity sensors collect real-time location humidity data for the corresponding areas; and the air velocity sensor integrated into the VAV terminal unit collects air velocity data from the air supply terminal of each unit. These environmental parameters provide the necessary input for assessing local comfort.

[0106] Once the energy control unit generates the first control command (triggering pre-adjustment of the HVAC set temperature), the data processing unit immediately activates additional monitoring processes. It continuously receives multi-dimensional data streams from the entire sensor network in real time, including updated personnel distribution data, location temperature data, location humidity data, and terminal wind speed data. The processing unit first identifies specific areas within the room where personnel are still present, based on the latest two-dimensional personnel location distribution map (from the status fusion module). These areas are defined as remaining personnel location areas. Since the two-dimensional location distribution map is based on pressure-sensing grid cells, a mapping relationship can be pre-established between the pressure-sensing grid cells and the areas defined in this embodiment. The remaining personnel location areas can be traced through the two-dimensional personnel location distribution map. For example, in a large open-plan office, the system might identify two areas where personnel are concentrated: the northwest corner and the southeast corner.

[0107] Next, the data processing unit uses a pre-defined sensor-area mapping relationship to precisely match the corresponding environmental parameter data for each identified remaining personnel location area. This mapping relationship is typically established during system deployment through spatial calibration to ensure that the temperature, humidity, and wind speed data for each area come from the nearest or most relevant sensors. For example, the data for the personnel area in the northwest corner is correlated with data collected by specific temperature and humidity sensors installed in that area, as well as the wind speed data collected by the VAV terminal wind speed sensor serving that area.

[0108] The core function of the data processing unit is to calculate the real-time local thermal comfort value for each remaining personnel location area using the PMV (Predicted Average Voting) model. The PMV model is an internationally recognized thermal comfort evaluation index that comprehensively considers factors such as temperature, humidity, wind speed, personnel activity levels, and clothing. In this solution, based on matched location temperature data, location humidity data, and terminal wind speed data, combined with preset standard personnel metabolic rates and clothing thermal resistance values ​​(usually set according to room usage), the system calculates a PMV value for each area. This value typically ranges from -3 (too cold) to +3 (too hot), with 0 representing neutral comfort. For example, a PMV value of 0.5 for an area indicates that people in that area may feel slightly warmer.

[0109] The system continuously compares the real-time PMV value of each area with a preset comfort threshold range (e.g., between -0.5 and +0.5). Once the PMV value of any area is detected to exceed this threshold range—whether in a positive direction (overheating) or a negative direction (overcooling)—the data processing unit immediately sends a compensation trigger signal to the energy control unit. Simultaneously, this signal carries the target variable air volume (VAV) terminal device identifier, clearly indicating the specific terminal device that needs adjustment. For example, if the PMV value in the southeast corner area reaches +0.6 (slightly exceeding the upper limit), the system will send a trigger signal and specify the VAV terminal ID serving that area.

[0110] While performing pre-adjustment of the temperature setpoint (i.e., phased linear temperature control), the energy control unit will simultaneously generate a dynamic compensation command if it receives a compensation trigger signal and target identifier. This command is specifically for the target variable air volume (VAV) terminal device, and its core requirement is to increase the airflow of the terminal device while maintaining the original set temperature adjustment process (e.g., continuing to increase the cooling setpoint temperature by 0.2°C per minute as planned). The increase in airflow is precisely calculated by the formula: ΔAirflow = k × |PMV Offset|. Wherein, PMV offset is the absolute difference between the real-time PMV value and the comfort threshold boundary value (e.g., +0.5 in the example above) (in this example, |0.6-0.5|=0.1); k is a preset proportionality coefficient, ranging from 10%-30% / 0.1 PMV units (e.g., k=20% / 0.1 PMV units). In this example, ΔAirflow = 20% × 0.1 = 2%, meaning that the airflow of the terminal device needs to be increased by 2%. This calculation ensures that the compensation is proportional to the degree of discomfort.

[0111] In practice during the pre-conditioning phase, for example, if the set temperature of a conference room is slowly increased at a rate of 0.2°C per minute (cooling mode), some people may experience slight discomfort due to the temperature rise or localized airflow changes. The system uses densely distributed sensors to detect in real-time that the PMV value in a certain area has risen to 0.6 (slight overheating) and immediately triggers a compensation mechanism. The VAV terminal serving that area increases the airflow by 2% within seconds (based on k=20%), increasing airflow to alleviate the stuffiness, while the overall room temperature increase process continues. The technical effect of this solution is a significant improvement in the user experience and robustness of the system. Without interrupting the overall energy-saving strategy (pre-conditioning), it eliminates potential discomfort caused by localized hot or cold spots through real-time compensation, ensuring that remaining occupants always perceive an acceptable thermal environment (local comfort maintained within the threshold) during their departure. The linear relationship between airflow compensation and PMV offset enables precise control, avoiding energy waste caused by over-compensation. Ultimately, this mechanism allows the system to more aggressively promote energy-saving temperature adjustments during the pre-conditioning period (such as allowing a larger ΔT0), while maintaining perceived comfort through local airflow compensation, thus achieving an effective balance between energy-saving goals and user satisfaction.

[0112] Furthermore, the response delay of the air volume adjustment does not exceed 15 seconds.

[0113] Here, setting a 15-second response delay limit ensures the timeliness of thermal comfort compensation. While the human body's perception of environmental changes has a certain lag, prolonged exposure to uncomfortable environments leads to significant negative feelings. The mandatory 15-second response time limit guarantees that within seconds to tens of seconds of people in a localized area beginning to feel slight discomfort (slightly exceeding the PMV limit), the system can intervene by increasing airflow (alleviating overheating) or decreasing airflow (alleviating overcooling), effectively preventing the accumulation and aggravation of discomfort. This rapid response is crucial for maintaining the immediate comfort perception of the remaining personnel, especially in the context of gradual changes in environmental parameters due to energy-saving pre-conditioning. It also enhances user trust and acceptance of the entire system, as users can intuitively feel the system's rapid and effective response to their comfort feedback. Ultimately, this requirement supports the effective operation of the compensation mechanism, enabling the system to reliably ensure local thermal environment quality while actively implementing energy-saving strategies (pre-conditioning and temperature control), achieving a dynamic balance between energy saving and comfort.

[0114] Furthermore, the data processing unit and the sensor network transmit real-time personnel distribution data and environmental parameters via RS-485 bus or Ethernet communication protocol.

[0115] In the system deployment, various sensors deployed in specific rooms (including wide-angle cameras, carpet-like pressure sensor arrays, temperature sensors, humidity sensors, and wind speed sensors integrated into the VAV terminal) constitute a sensor network. These sensors continuously generate key information including personnel distribution data (such as personnel location distribution maps and grid occupancy counts) and environmental parameters (such as location temperature, location humidity, and terminal wind speed). This data needs to be efficiently and reliably transmitted to the data processing unit for centralized processing and analysis.

[0116] Data transmission between the aforementioned sensor network and the data processing unit must be achieved through one of two specified communication protocols: RS-485 bus or Ethernet communication protocol. RS-485 is a mature serial communication bus standard, renowned for its excellent noise immunity, long transmission distance (typically within 1200m), and multi-point communication capability (supporting multiple devices connected to the same bus). Ethernet communication protocol, on the other hand, is a widely used local area network (LAN) technology standard that supports higher data transmission rates (such as 10Mbps, 100Mbps, 1Gbps, etc.) and provides reliable data packet transmission based on protocol stacks such as TCP / IP.

[0117] In actual system operation, such as in a medium-sized conference room scenario, sensor network nodes (e.g., ceiling-mounted wide-angle cameras, carpet pressure sensors, wall temperature and humidity sensors, and VAV terminal wind speed sensors) are connected to the communication line through their respective interfaces. If an RS-485 solution is used, these sensors may be connected to an RS-485 bus cable running through the room via an RS-485 conversion module, ultimately connecting to the RS-485 interface of the data processing unit. Sensor data is sent serially, either in turn or on demand, to the bus for reception by the data processing unit. If an Ethernet solution is used, each sensor or sensor group may be equipped with an Ethernet interface (e.g., RJ45), connected via a network cable to an Ethernet switch in the room, and then connected to the Ethernet interface of the data processing unit via the switch. Data is packaged into data frames and transmitted to the data processing unit via the Ethernet protocol. Both solutions ensure that real-time personnel distribution data generated by the sensors (e.g., grid occupancy counts updated per second, newly identified personnel locations) and environmental parameters (e.g., temperature values ​​sampled per second) are delivered to the data processing unit promptly and completely. The advantages of choosing RS-485 or Ethernet as the transmission medium lie in providing high reliability and industrial applicability. In complex building electrical environments, RS-485's excellent common-mode interference immunity effectively resists electromagnetic noise, ensuring signal transmission stability, and is particularly suitable for scenarios with a wide sensor distribution range, long wiring distances, and moderate bandwidth requirements. Ethernet, on the other hand, offers higher bandwidth and flexibility, suitable for situations requiring the transmission of large amounts of data (such as high-resolution images) or integration into existing enterprise network architectures. Both wired communication protocols avoid the signal interference, instability, or security risks that may exist with wireless transmission, ensuring the continuity and accuracy of core sensor data, and providing a solid data transmission foundation for the aforementioned complex personnel status analysis, trend prediction, and precise control. At the same time, the choice of protocol also provides deployment flexibility, allowing configuration based on the specific room size, existing infrastructure, and performance requirements.

[0118] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A smart building energy management and control system, characterized in that, include: Sensor networks are deployed in specific rooms to collect real-time data on the distribution of people. The data processing unit, which is connected to the sensor network, includes a real-time analysis module. The real-time analysis module calculates the current personnel density value of a specific room, the moving average value of the personnel density value within multiple consecutive sampling periods, and its dispersion based on real-time personnel distribution data. When the arithmetic mean of the dispersion within a predetermined continuous time window is detected to be lower than a first threshold and shows a continuous downward trend, it is determined that the specific room is about to enter a low personnel density state. An energy control unit is connected to a data processing unit and energy-consuming devices in the room. When it receives a judgment signal from the data processing unit indicating that a specific room is about to enter a low-personnel-density state, it generates a first control command. When the room's personnel density reaches a second threshold, it generates a second control command. The energy-consuming devices include a climate control system. The first control command is used to pre-adjust the climate control system of the specific room during the transition period from when the specific room is determined to be about to enter a low-personnel-density state until the personnel density of the specific room reaches a second threshold. The second control command is used to turn off the climate control system of the specific room; The pre-adjustment includes: linearly increasing the cooling temperature setpoint of the climate control system or linearly decreasing the heating temperature setpoint of the climate control system in stages according to the real-time rate of decrease of the current occupancy density in the room; the rate of linear increase or decrease is in the range of 0.1℃ to 0.5℃ per minute.

2. The intelligent building energy management system as described in claim 1, characterized in that, The sensor network includes a wide-angle camera deployed at a predetermined height above the ceiling in a specific room and a carpet-like pressure sensor array deployed at a predetermined location on the floor of the specific room. The wide-angle camera has a field of view of not less than 170 degrees and acquires image data including the outline of a person's head at a first sampling frequency. The carpet-like pressure sensor array consists of multiple pressure sensing areas arranged in a grid, with each grid cell having a side length ranging from 15cm to 30cm. The carpet-like pressure sensor array acquires foot pressure distribution data at a second sampling frequency. The data processing unit also includes a coordinate calibration module and a state fusion module; The coordinate calibration module, based on a predefined three-dimensional spatial coordinate mapping relationship of the room, transforms the head contour coordinates identified in the image data and the foot pressure center coordinates identified in the foot pressure distribution data into two-dimensional plane coordinates in the same room coordinate system, and pairs and associates the two-dimensional coordinates of the head contour and the two-dimensional coordinates of the foot pressure center at the same moment. The state fusion module calculates the dwell time of people in each pressure-sensing grid cell based on the successfully paired coordinate data. When the dwell time in a single pressure-sensing grid cell continuously exceeds a preset static determination threshold, it determines that there are stationary people in that grid cell. It aggregates all grid cells that have been determined to have stationary people to generate a two-dimensional location distribution map of people in the specific room. The personnel distribution status data includes the two-dimensional location distribution map of people and the real-time occupancy count of the pressure-sensing grid cells.

3. The intelligent building energy management system as described in claim 2, characterized in that, The real-time analysis module calculates the current person density value in the room, calculates the moving average of the person density value over a predetermined number of consecutive time sampling periods, and calculates the dispersion of the real-time person density value of each sampling point relative to the moving average over the predetermined number of consecutive time sampling periods, including the following methods: The real-time personnel density value at the current moment is calculated based on the ratio of the total number of pressure sensing grid cells occupied in the current sampling period to the total number of pressure sensing grid cells in the specific room. A weighted moving average is calculated for the real-time personnel density value sequence of N consecutive sampling periods, where N is an integer greater than or equal to 10 and less than or equal to 30. In the weighted moving average calculation, the real-time personnel density values ​​corresponding to the most recent m sampling periods are given higher weight coefficients, where m is an integer greater than or equal to 3 and less than or equal to 8, and the weight coefficients corresponding to the sampling periods closer to the current time are larger. Calculate the variance between the real-time population density value of each sampling point and its corresponding weighted moving average value within the N consecutive sampling periods, and use the variance as a quantitative indicator of dispersion.

4. The intelligent building energy management system as described in claim 3, characterized in that, The specific process by which the real-time analysis module determines that a particular room is about to enter a low-personnel-density state includes: Linear regression analysis was performed on the real-time personnel density value sequence of the N consecutive sampling periods to obtain the slope value of personnel density changing with time; When preset conditions are met simultaneously, it is determined that the specific room is about to enter a low-personnel-density state. The preset conditions include: a) the variance is lower than a preset variance threshold for K consecutive sampling periods, where K is an integer greater than or equal to 3 and less than or equal to 10; b) the slope value is negative; c) the absolute value of the slope value is greater than a preset slope threshold for K consecutive sampling periods; wherein, the variance threshold ranges from 0.5 to 1.5, and the slope threshold ranges from a decrease of 0.05 to 0.2 density units per minute.

5. The intelligent building energy management system as described in claim 4, characterized in that, The pre-adjustment process of the energy control unit specifically includes: The real-time rate of decrease in personnel density, calculated in real time by the data processing unit, is obtained. The real-time rate of decrease in personnel density is the absolute value of the slope value obtained by linear regression analysis. The real-time rate of decrease in personnel density is input into a predefined adjustment coefficient calculation function, and a first adjustment coefficient is output. The adjustment coefficient calculation function satisfies the following: when the real-time rate of decrease in personnel density is within the range of a preset lower threshold to an upper threshold, the first adjustment coefficient is directly proportional to the real-time rate of decrease in personnel density, and the proportionality constant ranges from 0.5 to 2.

0. Based on the first adjustment coefficient α, calculate the target adjustment range ΔT of the climate control system set temperature during the transition period: ΔT = α × ΔT0, where ΔT0 is the basic adjustment range, which ranges from 1.0℃ to 3.0℃; The transition period is divided into two consecutive adjustment sub-stages. In the first adjustment sub-stage, the cooling temperature setpoint or heating temperature setpoint of the climate control system is adjusted in the direction of the target adjustment range at a first temperature change rate. In the second adjustment sub-stage, the setpoint is further adjusted in the direction of the target adjustment range at a second temperature change rate. Wherein, the first temperature change rate is less than the second temperature change rate; the first temperature change rate ranges from 0.1℃ to 0.3℃ per minute, and the second temperature change rate ranges from 0.2℃ to 0.5℃ per minute.

6. The intelligent building energy management system as described in claim 5, characterized in that, The sensor network also includes multiple temperature sensors, multiple humidity sensors, and wind speed sensors of several variable air volume terminal devices integrated into a climate control system, all deployed in a specific room. The multiple temperature sensors are used to collect location temperature data, the multiple humidity sensors are used to collect location humidity data, and the wind speed sensors are used to collect terminal wind speed data. The data processing unit has a preset sensor-area mapping relationship: the specific room is divided into several areas, each area corresponds to a temperature sensor, a humidity sensor and a variable air volume terminal device. Each area is numbered, and the temperature sensor, humidity sensor, wind speed sensor and variable air volume terminal device in the area are given the same unique identifier. The identifiers of the temperature sensor, humidity sensor, wind speed sensor and variable air volume terminal device in each area are associated with the number of the area. After the energy control unit generates the first control command, the data processing unit receives in real time personnel distribution status data, location temperature data, location humidity data, and terminal wind speed data collected from the sensor network. Based on the personnel distribution status data, it identifies the remaining personnel location areas in the current room. According to the preset sensor-area mapping relationship, it matches the corresponding location temperature data, location humidity data, and terminal wind speed data for each remaining personnel location area. Using the PMV model, it calculates the real-time local thermal comfort value of each remaining personnel location area based on the matched location temperature data, location humidity data, and terminal wind speed data. When any real-time local thermal comfort value exceeds the preset comfort threshold range, it sends a compensation trigger signal and the corresponding target variable air volume terminal device identifier to the energy control unit. While the energy control unit outputs the first control command for pre-adjustment, if it receives the compensation trigger signal and the target variable air volume terminal device identifier, it generates a dynamic compensation command. The dynamic compensation command directionally controls the target variable air volume terminal device to increase its air volume while maintaining the set temperature adjustment process. The increase in air volume is: Δ air volume = k × |PMV offset|, where k is a proportional coefficient and PMV offset is the difference between the real-time local thermal comfort value and the comfort threshold boundary value.

7. The intelligent building energy management system as described in claim 6, characterized in that, The response delay for the air volume adjustment shall not exceed 15 seconds.

8. The intelligent building energy management system as described in claim 1, characterized in that, The data processing unit and the sensor network transmit real-time personnel distribution data and environmental parameters via RS-485 bus or Ethernet communication protocol.

Images